loicsapone commited on
Commit
3dbf962
·
verified ·
1 Parent(s): 26fe8dc

Update model card for javascript

Browse files
Files changed (1) hide show
  1. README.md +105 -0
README.md ADDED
@@ -0,0 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - code-review
5
+ - javascript
6
+ - mlx
7
+ - gguf
8
+ - qwen2.5-coder
9
+ base_model: Qwen/Qwen2.5-Coder-1.5B-Instruct
10
+ ---
11
+
12
+ # AI Code Review Model - Javascript
13
+
14
+ This is a fine-tuned code review model specialized for **Javascript** code analysis.
15
+
16
+ ## Model Details
17
+
18
+ - **Base Model**: Qwen/Qwen2.5-Coder-1.5B-Instruct
19
+ - **Training Method**: LoRA fine-tuning with MLX
20
+ - **Format**: GGUF (Q4_K_M quantization)
21
+ - **Target Language**: Javascript
22
+ - **Purpose**: Automated code review for CI/CD pipelines
23
+
24
+ ## Usage
25
+
26
+ ### Docker (Recommended)
27
+
28
+ ```bash
29
+ docker pull ghcr.io/iq2i/ai-code-review:javascript-latest
30
+ docker run --rm -v $(pwd):/workspace ghcr.io/iq2i/ai-code-review:javascript-latest /workspace/src
31
+ ```
32
+
33
+ ### llama.cpp
34
+
35
+ ```bash
36
+ # Download the model
37
+ wget https://huggingface.co/loicsapone/ai-code-review-javascript/resolve/main/model-Q4_K_M.gguf
38
+
39
+ # Run inference
40
+ ./llama-cli -m model-Q4_K_M.gguf -p "Review this code: ..."
41
+ ```
42
+
43
+ ### Python (llama-cpp-python)
44
+
45
+ ```python
46
+ from llama_cpp import Llama
47
+
48
+ llm = Llama(model_path="model-Q4_K_M.gguf")
49
+ output = llm("Review this code: ...", max_tokens=512)
50
+ print(output)
51
+ ```
52
+
53
+ ## Output Format
54
+
55
+ The model outputs JSON structured code reviews:
56
+
57
+ ```json
58
+ {
59
+ "summary": "Brief overview of code quality",
60
+ "score": 8,
61
+ "issues": [
62
+ {
63
+ "type": "bug",
64
+ "severity": "medium",
65
+ "line": 42,
66
+ "description": "Potential null pointer",
67
+ "suggestion": "Add null check"
68
+ }
69
+ ],
70
+ "positive_points": [
71
+ "Good error handling",
72
+ "Clear variable names"
73
+ ]
74
+ }
75
+ ```
76
+
77
+ ## Training
78
+
79
+ This model was trained on curated Javascript code review examples using:
80
+ - MLX framework for Apple Silicon acceleration
81
+ - LoRA adapters (r=8, alpha=16)
82
+ - Custom dataset of real-world code issues
83
+
84
+ For training details, see the [GitHub repository](https://github.com/iq2i/ai-code-review).
85
+
86
+ ## Limitations
87
+
88
+ - Optimized for Javascript syntax and best practices
89
+ - May not catch all edge cases or security vulnerabilities
90
+ - Should be used as a supplementary tool, not a replacement for human review
91
+
92
+ ## License
93
+
94
+ Apache 2.0
95
+
96
+ ## Citation
97
+
98
+ ```bibtex
99
+ @software{ai_code_review_javascript,
100
+ title = {AI Code Review Model for Javascript},
101
+ author = {IQ2i Team},
102
+ year = {2025},
103
+ url = {https://github.com/iq2i/ai-code-review}
104
+ }
105
+ ```